Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
Operating Systems Software Linux

Linux Kernel v2.6.23 Released 346 writes "After 3 months, Linus has released Linux 2.6.23. This version includes the new and shiny CFS process scheduler, a simpler read-ahead mechanism, the lguest 'Linux-on-Linux' paravirtualization hypervisor, XEN guest support, KVM smp guest support, and variable process argument length. SLUB is now the default slab allocator, there's SELinux protection for exploiting null dereferences using mmap, XFS and ext4 improvements, PPP over L2TP support. Also the 'lumpy' reclaim algorithm, a userspace driver framework, the O_CLOEXEC file descriptor flag, splice improvements, a new fallocate() syscall, lock statistics, support for multiqueue network devices, various new drivers, and many other minor features and fixes. See the changelog for details."
This discussion has been archived. No new comments can be posted.

Linux Kernel v2.6.23 Released

Comments Filter:
  • by speaker of the truth ( 1112181 ) on Tuesday October 09, 2007 @11:31PM (#20921697)
    I RTFA and it didn't mention whether or not it was released under GPL v2 or v3. Does anyone know?
  • I love my Thinkpad (Score:3, Interesting)

    by wellingj ( 1030460 ) on Tuesday October 09, 2007 @11:37PM (#20921761)
    thinkpad-acpi: enable more hotkeys, add input device support to hotkey subdriver

  • Methinks... (Score:5, Interesting)

    by Keyper7 ( 1160079 ) on Tuesday October 09, 2007 @11:42PM (#20921803)
    ...the extra flavor that makes this release a little bit more headline-worthy than usual is probably the whole controversy involving the Completely Fair Scheduler. Between Con Kolivas leaving kernel development, the Really Fair Scheduler flamewar and almost ten release candidates, the whole 2.6.23 development was some kind of geek soap opera.
  • by Myria ( 562655 ) on Tuesday October 09, 2007 @11:54PM (#20921911)

    In multi-threaded code (or more correctly: all code using clone() with CLONE_FILES) there's a race when exec'ing (see commit link for details). In some applications this can happen frequently. Take a web browser. One thread opens a file and another thread starts, say, an external PDF viewer. The result can even be a security issue if that open file descriptor refers to a sensitive file and the external program can somehow be tricked into using that descriptor. 2.6.23 includes the O_CLOEXEC ("close-on-exec") fd flag on open() and recvmsg() to avoid this problem.

    Yes, this is a good thing. However, they seem to have missed some: sockets and pipes. Sockets are not close-on-exec by default, so you may pass a sensitive socket to a child.

    Windows NT has the same problem: sockets are inheritable by default until you call SetHandleInformation to disable inheritance. Other handles' inheritability is selected at open/create time.

    Luckily, there is a workaround for it, if not pretty: use a reader/writer lock with opening handles as writers and forks as readers.

    By the way, the linked changelog on has a bad link for the "recommended LWN article".

    For the SELinux thing against null pointer attacks, won't that break DOSemu?
  • by tjstork ( 137384 ) <todd,bandrowsky&gmail,com> on Wednesday October 10, 2007 @12:03AM (#20922023) Homepage Journal
    fallocate() is a new system call which will allow applications to preallocate space to any file(s) in a file system. Applications can get a guarantee of space for particular file(s) - even if later the system becomes full

    I was about to go and make fun of Linux for creating a feature that's been around in Windows for quite a while - take your pick of SetFilePointer or sparse files. Yes, yes, I understand that reserving space for a file is not the same as growing it and not using that space. Twas meant to be a troll....But, it turns out that a bit of googling reveals that sparse files under Windows are not all that they are cracked up to be: []

  • by CandyMan ( 15493 ) <> on Wednesday October 10, 2007 @12:45AM (#20922415) Homepage
    Sun will probably publish Solaris under the GPL v3, so everyone will have a choice of tree free kernels: *BSD under BSD, Linux under GPL v2 and Solaris under V3. I think there is a fair chances that some developers might want to dual-license their code from now on. I am thinking of someone publishing their new filesystem code under both GPL versions so both projects can use it.

    Relicensing existing code might be too strenuous, but if many developers decide to follow this dual-licensing approach, the relicensing of Linux may be made easier by module replacement, as old GPL v2 code is swapped out for new "either GPL v2 or v3" dual-licensed code coming in.

    In any case, this is highly speculative, and as much as I would like Linux to be under the GPL v2 (I think tivoization sucks), if its authors don't care about it as much as we do, I don't feel inclined to raise a stink. Or maybe I am inclined to raise it against tivoizers, but not against developers themselves. We can still use Linux, and I for one thank our kernel developer overlords for their good job working for all of us.

    (Note: I know there are several BSD kernels, but that's true also of Linux: there are several forks for different uses and profiles).
  • Userspace drivers? (Score:3, Interesting)

    by NereusRen ( 811533 ) on Wednesday October 10, 2007 @01:00AM (#20922545)

    a userspace driver framework
    I have a question for someone better-informed than myself: Does this mean we are a step closer to not having to recompile nvidia's video drivers after installing a new kernel?
  • by ubiquitin ( 28396 ) * on Wednesday October 10, 2007 @01:07AM (#20922607) Homepage Journal
    An exploit with feature-complete proof of concept [] was released for x86_64 linux kernel ia32syscall emulation by cliph at isec in Poland. Exploit code was wildly popular on milw0rm [], indicating that this local exploit has lots of potential.
  • by QuantumG ( 50515 ) <> on Wednesday October 10, 2007 @01:09AM (#20922619) Homepage Journal
    Short answer: no.

    Long answer: if NVIDIA ever makes open source drivers, they will almost definitely be kernel space drivers. Apparently this is in the works, same with ATI, but I'll believe it when it happens. It would be possible for some bored hacker to take the NVIDIA binary blobs and make a userspace driver from them. This driver could be legally distributed with the NVIDIA binary blobs (probably). And yes, this would mean that recompiling the drivers for a new kernel would not be necessary.. and it would also mean that the kernel wouldn't be "tainted" by using this driver (maybe).

    I, personally, think the stability and security advantages of running binary blobs in userspace drivers outweighs the possible performance hit (no-one has measured the performance hit, yet), so it's a good idea. But, ya know, I've got some other stuff to do...

  • by setagllib ( 753300 ) on Wednesday October 10, 2007 @01:46AM (#20922841)
    It's called a network effect. Linux improves, and gets more users, some of whom are developers, who improve Linux. It just keeps growing with every cycle.

    Proprietary operating systems can't compete because they're closed. The best an innovative user/developer can do is fire off feedback asking for a feature, and it'll be implemented wrong anyway, and then released 3 years later in the next major version.

    Even more impressive is that this is the *stable* kernel branch that's growing so fast. The -mm experimental branch has gone right off the hook, to the point Andrew is complaining the development doesn't scale any more with only him at the helm.

    For those who want a more conservative choice for servers, there's always something like FreeBSD. It's nice to have choice and interoperability. FreeBSD is more compatible with Linux than Windows XP is compatible with Windows Vista. If you don't believe me, consider that at least FreeBSD and Linux have a lot of standards (APIs, file formats, layouts, etc) in common.
  • by dbIII ( 701233 ) on Wednesday October 10, 2007 @02:12AM (#20922979)
    You are correct - that is a really bad analogy. A lot of people just drop the new kernel into whatever distro they have if their problem bit of hardware has a better driver or there's a speed improvement somewhere. Remember even a 1% speed improvement cuts over an hour off the runtime of a weeklong numerical processing job.
  • by tjstork ( 137384 ) <todd,bandrowsky&gmail,com> on Wednesday October 10, 2007 @02:46AM (#20923195) Homepage Journal
    Is anyone else noticing the almost exponential rise in the rate at which new features are being added to the kernel? Linux major release anouncements would dwarf similar anouncements by 'competing' operating systems.

    I think they write out every little thing they did, designed to more impress than really say oh wow, big new features. Microsoft major releases go in circles, but they do some pretty big stuff. Let's see, starting in NT4, they put the graphics drivers into the kernel, then a few releases later, they moved them out. Then they shifted the whole driver model around a few times. Then they put http protocol into the kernel, then they put the sound drivers outside of the kernel and probably down the road, something will inspire them to move http out of the kernel and put the sound drivers back into the kernel space. And, some of the features they've added along the way include incremental improvements to kernel queues, and, like Linux, MS seems to always be searching for a better scheduler.
  • bloat (Score:1, Interesting)

    by m2943 ( 1140797 ) on Wednesday October 10, 2007 @04:08AM (#20923591)
    Going over the list, there is little there I wanted and nothing I needed. I think the Linux kernel is getting more and more bloated.
  • Dreamcast support (Score:3, Interesting)

    by 00_NOP ( 559413 ) on Wednesday October 10, 2007 @06:10AM (#20924151) Homepage
    The new kernel also includes ALSA support for the Dreamcast sound device (for the first time - an out of mainline OSS driver did/does exist for 2.4).

    More dreamcast support is on the way - expect some more stuff in 2.6.24 and 2.6.25 and I (the author of the code) would love to hear from willing testers, etc
  • by Hal_Porter ( 817932 ) on Wednesday October 10, 2007 @06:24AM (#20924223)

    Don't use "meme" as a euphemism for "cliché".
    Cliches seem like an excellent example of memes actually.

    They have heredity - the actual text of the cliche. E.g. "In Soviet Russia X verb Y", or "In Korea only old people do X".

    They are subject to natural selection as popular memes will replicate faster by definition.

    They have mutations - random(ish) changes, typos or non sequitur that add humour. They even have sexual repoduction since memes can be combined for humorous effect. E.g. in a story about dogs attacking people in North Korea I could quip "In Soviet Korea, dog eat old people!" combining two memes. Both effects are used to avoid an analogue of Muller's ratchet [] where a stale meme is no longer funny and thus is not copied.

    They are also highly virulent to the point where they can take over message boards completely.
  • by DocDJ ( 530740 ) on Wednesday October 10, 2007 @07:13AM (#20924461)
    That's no cliché. It's a snowclone [].

The only possible interpretation of any research whatever in the `social sciences' is: some do, some don't. -- Ernest Rutherford